Impulsivity and Inhibitory Control in Normal Development and Childhood Psychopathology

Russell Schachar Gordon D. Logan Department of Psychiatry University of Illinois The Hospital for Sick Children Toronto, Ontario, Canada

Two experiments investigated the development and pathology of inhibitory control in children. Inhibitory control was investigated with the stop-signal paradigm, which is based on a formal theory of inhibition and directly measures the mechanism of inhibition. The ability to inhibit developed little after Grade 2, but subjects with attention deficit disorder with hyperactivity (ADDH) showed deficient inhibitory control. Their deficient inhibitory control was attributable to the subgroup of ADDH subjects with pervasive hyperactivity who had a more severe inhibitory deficit than did the situational hyperactive subgroup, the normal group, and the pathological controls. These studies reflect the utility of the stop-signal paradigm as a measure of inhibitory control.

The concept of inhibitory control is central in theories of measure of inhibition exists. Instead, impulsivity is defined op- child development and in the definition and explanation of erationally (e.g, calling out in class or demonstrating an inabil- psychopathological disorders of childhood—in particular, of ity to plan performance on a maze task). Most frequently, im- attention deficit disorder with hyperactivity (ADDH; Douglas, pulsivity and inhibitory control are studied with laboratory 1983; Kogan, 1983; Milich & Kramer, 1984). Inhibitory control tasks that involve response uncertainty, careful planning, and is one of several processes that perform the executive functions slow responding. The most common measure is the Matching of the cognitive system. These functions determine how various Familiar Figures Task (MFFT; Kagan, Rosman, Day, Albert, & mental processes (e.g., encoding, recognition, retrieval) will Phillips, 1964). This task requires subjects to search a number work together in the performance of a task. Children need exec- of similar pictures for one that matches a criterion picture ex- utive control to choose, construct, execute, and maintain opti- actly. Fast and inaccurate performance on the MFFT is as- mal strategies for performing a task, as well as to inhibit strate- sumed to indicate impulsivity or lack of cognitive control over gies that become inappropriate when goals or task demands the execution of the response; the subject is believed to be un- change or errors occur (Logan, 1985). Deficient inhibitory con- able to delay a response in the course of analyzing the stimuli trol is revealed by impulsive behaviors such as responding be- and searching for the correct alternative. fore the task is understood, answering before sufficient infor- Studies using the MFFT indicate that as normal children mation is available, allowing attention to be captured by irrele- grow older, they develop longer response latencies and greater vant stimuli (i.e, distractibility), or failing to correct obviously accuracy (Salkind & Wright, 1977). In contrast, children with inappropriate responses. Poorly developed inhibitory control clinical disorders characterized by apparent impulsivity, such might account for impulsive behaviors of younger children and as hyperactive children (attention deficit disorder, ADDH), re- of children with various types of psychopathology. spond more quickly and make more errors (see Campbell, Despite the importance of inhibitory control in theories of Douglas, & Morgenstern, 1971). These differences are inter- child development and psychopathology, no widely accepted preted as evidence of deficient impulse control in hyperactive and younger children. Impulsive performance on the MFFT and other tasks such as the continuous performance task is a We would like to acknowledge the contributions of Diane Chajczyk, major argument for the hypothesis that deficient impulse con- Pat Fulford, Haraldene Phair, and Rosemary Tannock in the conduct trol is one of the basic problems of hyperactive children (Doug- of this study and the support of the Ontario Mental Health, Ruth Schwartz, and Canadian Psychiatric Research Foundations. The man- las, 1983; Douglas & Peters, 1979). uscript was prepared with the assistance of the Medical Publications Performance on these tasks, however, depends on factors Department, The Hospital for Sick Children, Toronto, Ontario, Can- other than cognitive impulsivity. For example, MFFT perfor- ada. mance varies with IQ (Milich & Kramer, 1984); search strategy Correspondence concerning this article should be addressed to Rus- (Ault, Crawford, & Jeffrey, 1972); metacognitive awareness of sell Schachar, Department of Psychiatry, The Hospital for Sick Chil- appropriateness of inhibiting the response until all variants dren, 555 University Avenue, Toronto, Ontario, Canada, M5G 1X8 or to Gordon D. Logan, Department of Psychology, University of Illinois, have been compared (Brown, Bransford, Ferrara, & Campione, 603 East Daniel Street, Champaign, Illinois 61820. Electronic mail 1983); or decision criteria, motivation, or attentional capacity may be sent to g-logan @ h. psych, uiuc. or to attn @ sick kids, utoronto. (Kahneman, 1973). Consequently, the impulsivity that the MFFT and similar tasks reveal in younger and hyperactive

710 IMPULSIVITY AND INHIBITORY CONTROL 711 children could be accounted for by deficits in one or several times of these processes are assumed to vary randomly, the processes. Tasks like the MFFT may help distinguish between outcome of the race is a matter of probability. Thus, in the individuals or groups but do not clarify the underlying psycho- model, the main dependent variable is the probability of inhibi- logical processes that produce impulsivity. tion. In addition to difficulties in measuring impulsivity, clinical The main independent variable is the interval between the studies have been hampered by a lack of diagnostic precision. stop signal and the primary-task response (i.e, the stop-signal Many studies of impulsivity in hyperactive children have used delay). The stop-signal delay handicaps the race, biasing it in samples of children exhibiting a variety of comorbid disorders favor of one process or the other. If the signal occurs early or have not included pathological controls. Hyperactive chil- enough (i.e, well before the primary-task stimulus), subjects dren typically present with a variety of other disturbances, such will inhibit every time. If it occurs late enough (i.e, after the as conduct and learning disorders. To determine whether hyper- response to the primary task), subjects will never inhibit. Be- activity rather than one or all of these correlated disorders is tween these extremes, the probability of inhibition diminishes associated with cognitive impulsivity, hyperactive children gradually from 1.0 to 0.0. Plotting the probability of inhibition must be compared with children who have one of these dis- against stop-signal delay generates an inhibition function orders, as well as with normal children. whose shape depends on the distribution of primary-task reac- In this study, we investigated the development and pathology tion times and the distribution of internal reaction times to the of inhibitory control through the use of a new stop-signal para- stop signal (the stop-signal reaction time, or SSRT; for a formal digm that allows direct assessment of inhibitory control. In the derivation, see Logan & Cowan, 1984). first experiment, we focused on the development of inhibitory Differences in the efficiency of the inhibition process will control in normal children as measured in the stop-signal para- affect the shape of the inhibition function. In general, the better digm. In the second experiment, we compared the inhibitory the process, the higher and steeper the function. If the inhibi- control of normal children with that of children who have a tion mechanism were never triggered in stop-signal trials, the psychopathological condition characterized clinically by poor probability of inhibition would be 0.0 at all delays, producing a impulse control (ADDH). To determine whether a deficit in flat, low inhibition function. If the mechanism were triggered inhibitory control is specific to ADDH, we compared the per- in only some stop-signal trials, the function would be steeper formance of these children with that of children who have emo- and higher, but not as steep and high as a normal one. The speed tional, learning, and conduct disorders. of the inhibition mechanism also determines the shape of the inhibition function. A slower mechanism will lose the race against the primary-task process more often than a faster one, The Stop-Signal Paradigm producing a lower probability of inhibition. Extended over all The stop-signal paradigm is a laboratory analog of a situation delays, it will produce a lower and flatter inhibition function. If requiring inhibitory control that distinguishes stimuli that the inhibition mechanism of younger children or children with elicit impulsive behavior (primary-task stimuli) from those that a clinical condition affecting impulse control is less likely to be inhibit .it (stop-signal stimuli). Subjects engaged in a primary triggered, is slower, or is less consistent, their inhibition func- task (e.g., forced-choice letter discrimination) are presented tions will be flatter and lower than those of older and normal with an occasional stop-signal stimulus (e.g., a tone) instructing children.' them to inhibit their response to the primary-task stimulus. According to the race model, the shape of the inhibition Unlike methods usually used in studies of childhood impulsiv- function also depends on the mean and variability of the pri- ity (e.g, Kagan et al, 1964), the stop-signal paradigm is based mary-task processes. If inhibition functions are linear, varia- on a theory of inhibition (Logan & Cowan, 1984; Logan, tion in means will affect the intercept of the inhibition function Cowan, & Davis, 1984) and directly measures the mechanism and variation in standard deviations will affect slope. In re- of inhibition. In the real world, the stimuli that elicit impulsive search with adults, these effects have been small (Logan et al, behavior and those that inhibit it sometimes overlap (e.g, when 1984). However, we expect greater differences among children the deviance or inappropriateness of the response is the signal because response times are longer and more variable in to inhibit it). We attempted to apply this paradigm only with younger and disturbed children than in older and undisturbed clearly discriminable "go" and "stop" stimuli to establish the children (Peloquin & Klorman, 1986). basic effects of development and psychopathology. Differences in the mean and the variability of the stopping Childhood impulsivity may be associated with deficiency of and primary-task reaction times among groups may or may not inhibitory control in two ways: The stop-signal stimulus either be sufficient to account for observed intergroup differences in may fail to trigger the stop-signal response or may trigger a inhibition functions. The model of Logan and Cowan (1984) response that is slower or more variable than normal (see Luce, 1986). 1 Variability in the speed of the inhibition process also affects the steepness of the function. According to the race model, the inhibition Inhibition Functions function can be viewed as a probability distribution: The more variable the distribution, the flatter the function. The race model asserts In the model of Logan and Cowan (1984), response inhibi- that the variance of the inhibition function is the sum of the variance tion depends on a race between the primary-task processes and of the SSRT and the primary-task reaction times. Thus, a deficiency the stop-signal processes. If the former win, a response occurs; that increases the variance of the SSRT will flatten the inhibition if the latter win, the response is inhibited. Because the finishing function (see Logan & Cowan, 1984). 712 RUSSELL SCHACHAR AND GORDON D. LOGAN provides methods of correcting inhibition functions for differ- Experiment 1: Normal Development ences in stopping and primary-task reaction times that can bring disparate inhibition functions into alignment. Method Differences in mean primary-task reaction time can be Subjects. Thirty-six boys (12 from each of Grades 2,4, and 6) and 12 corrected by setting the stop-signal presentation relative to the adult volunteers (research assistants, students) were subjects in this mean primary-task reaction time, as we have in our experi- study. The children were recruited from three schools through a two- ments (see Setting stop-signal delay for details). If inhibition stage procedure. First, teachers selected children who they believed functions are not aligned by this method of correcting for dif- were of average intelligence and without learning or behavioral prob- ferences in the mean primary-task reaction time, then group lems. Second, the absence of learning or behavior problems was con- differences may be a result of differences in the stop-signal firmed through behavior ratings completed by the child's teacher and parents and through measures of intelligence and academic attain- reaction time or of variability in the primary-task reaction ment. Informed consent for the experiment was obtained from the times. In order to account for differences in stop-signal reac- children's parents, and assent was obtained from each subject. Parents tion time, the probability of inhibition can be plotted as a func- and teachers completed (a) the appropriate version of the Rutter Behav- tion of MRT-delay-SSRT. If the functions are still not aligned, ior Rating Scale (Form A for parents and Form B for teachers; Rutter, they should be plotted as a function of ZRFT (Logan et al, 1967; Rutter, Tizard, & Whitmore, 1970), (b) the SNAP questionnaire 1984): (Pelham, Atkins, & Murphy, 1981), and (c) the Abbreviated Conners Teacher Rating Scales (ACTRS; Conners, 1973). To be included in the MRT - delay - SSRT sample, children had to obtain scores within the normal range on both ZRFT = parent and teacher questionnaires. More specifically, children had to SDRT obtain the following scores: (a) less than 9 on the Rutter-B and less than 13 on the Rutter-A rating scales (scores greater than these indicate where ZRFT is the relative finishingtime s of the stopping and psychiatric disturbance); (b) less than 4 on the Inattentive subscale, less primary-task processes, expressed as a Z score; MRT is the than 4 on the Impulsive subscale, and less than 3 on the Hyperactive mean primary-task response time; delay is the interval between subscale of the SNAP questionnaire (higher scores, obtained by 5% of the stop-signal stimulus and the primary-task response; SSRT 10-year-old boys, indicate a diagnosis of hyperactivity; Pelham et al, is the estimated stop-signal response time; and SDRT is the 1981); (c) less than 15 on the ACTRS (scores above 15 are indicative of a standard deviation of the primary-task response times (see Lo- diagnosis of hyperactivity; Goyette, Conners, & Ulrich, 1978). 2 gan & Cowan, 1984; Logan et al, 1984). If the inhibition func- To confirm that each child was of at least average intelligence, we tions from different age or diagnostic groups cannot be aligned administered two subtests (Vocabulary and Block Design) of the by plotting them as a function of ZRFT, then we conclude that Wechsler Intelligence Scale for Children-Revised (WISC-R) the shallower functions represent deficiencies in the executive (Wechsler, 1974) that correlate most highly with the full-scale score. process of inhibition; either the inhibitory process has more The Wide Range Achievement Test (WRAT-R; Jastak & Wilkinson, variability, or it is triggered less often in the groups with the 1984) was used to estimate level of academic attainment. To be in- shallower slope. cluded in the study, children had to have full-scale IQ scores of 80 or more and score above the 25th percentile on the WRAT-R Reading, Spelling, and Arithmetic tests. The mean age of the three groups of students was 95,118, and 145 months, respectively, and their mean IQs Estimates of Stop-Signal Reaction Time did not differ, F(2, 33) = 0.6, ns. Information on the age and IQ of the adult volunteers was not collected. Subjects were not paid for their The race model suggests a way to estimate the time needed to participation. respond to the stop signal at any given delay. Although the Stopping task. The experimenter-paced stopping task, developed SSRT is not observable directly, it may be defined with respect by Logan and colleagues (Logan et al, 1984; Logan & Cowan, 1984), to the distribution of primary-task reaction time (RT). SSRT examines the ability to inhibit responses to a choice reaction-time task in which both speed and accuracy are emphasized. The primary-task can be treated as a constant that appears as a point on the stimuli were the uppercase letters X and O, presented by an Apple lie primary-task RT distribution. Reactions occurring before this computer connected to a specialized Cognitive Testing Station (CTS; point will not be inhibited, but those coming after it will be Digitry Company, Inc., Maine, 1984), which allowed direct and precise inhibited. The proportion of the distribution preceding this control of the stimulus presentation, as well as the collection of re- point represents the probability of responding despite a stop sponse times, with millisecond timing. Each letter, presented one at a signal; the proportion following it represents the probability of time in the center of the screen, was 2 mm wide and 5 mm high and, inhibition. The point may be estimated from the percentile when viewed at a distance of 40 cm, subtended 0.29° X 0.72° of visual value of the primary-task RT distribution that corresponds to angle. The stop signal was a 1 -kHz tone (beep) generated and presented one minus the probability of inhibition. This RT represents the by the computer. Half of the stop signals occurred with an X and half completion of the stopping process relative to the primary-task with an O. The details of the stop-signal delays and the manner in stimulus. To express the completion of the stopping process which they were generated are presented below. relative to the stop-signal stimulus (i.e, to calculate SSRT), the Each trial began with a fixation point illuminated for 500 ms. It was followed by the letter for that trial, displayed for 1 s and then extin- stop-signal delay must be subtracted (for a formal derivation, guished. The screen remained blank for an interval of 1.5 s. Thus, each see Logan & Cowan, 1984). \bunger and disturbed children trial included a period of 2.5 s in which the subject could respond. The would be expected to show longer SSRTs than older and normal children. Longer SSRTs in disturbed children compared with those in normal children of similar age and IQ can be viewed as 2 It is not possible to correct for SDRT without prior adjustment for evidence of a deficiency in inhibitory control. SSRT (Logan et al, 1984). IMPULSIVITY AND INHIBITORY CONTROL 713 subject responded by pressing one of two keys on a response box con- the inhibition functions plotted as a function of ZRFT to determine nected to the CTS hardware. Mapping of letters onto the keys was the extent to which primary-task variability rather than differences in counterbalanced across subjects. inhibitory control accounted for observed differences in the slope of The test trials were presented in blocks of 48 trials; the blocks were the inhibition function. The slopes of the regression lines for each organized in three groups of three blocks (total of 432 test trials), with a subject were entered into a univariate ANOVA. short break between the groups. The two stimuli occurred equally For other variables, such as mean nonresponses, age groups were often in each block, and each stop-signal delay occurred equally often compared by univariate ANOVAS. For all measures in which the overall with each letter. Stop signals were presented in 25% of the trials (108 F value was significant, post hoc Newman-Keuls tests (Winer, 1971) trials), occurring equally often at each of six stop-signal delays, so that a were conducted to locate differences between age groups. Finally, the total of 18 stop signals occurred at each stop-signal delay. The sequence association of age and IQ with these indices of performance across all of primary-task stimuli, stop signals, and stop-signal delays was ran- subjects was assessed through correlation analysis. dom. The trials lasted from 35 to 45 min depending on the length of breaks. Results Setting stop-signal delay. To account for differences between subjects in primary-task RT and strategy (e.g, a subject may have held back Primary-task processes. Age groups differed in perfor- a response in an attempt to increase the probability of inhibiting), mance on the primary task (Table 1). Older subjects responded stop-signal delay was denned as the interval between the stop-signal more quickly and with less variability; differences were particu- stimulus and the subject's expected primary-task RT. Specifically, larly marked between the adult and Grade 6 groups and the MRT was measured in each block of trials, and stop-signal stimuli younger groups. The groups did not differ in mean percentages were presented at delays defined relative to MRT, tracking changes in of errors or of nonresponses. MRT over time. Logan et al. (1984) showed that various methods of setting the delay produced equivalent results, but tracking MRT pro- Stopping processes. Table 2 presents the mean probability of duced better inhibition functions. Tracking makes it easier to capture inhibiting a response to the primary task at each stop-signal the sloping section of the inhibition function between the high and low delay for each age group. Older subjects inhibited a greater pro- asymptotes, which is the most informative part theoretically. portion of responses over all delays, but the differences were Before the 432 test trials, a practice block of 48 trials was presented not significant, F(3, 44) = 1.9. to estimate the mean choice response time. During this practice block, As predicted by the model, the probability of inhibition in- stop signals were presented but subjects were not told their purpose. creased significantly as stop-signal delay increased, F(5,220) = They were presented arbitrarily 500, 400, 300, 200,100, and 0 ms be- 126.2, p < .001 (Table 2 and Figure 1). All age groups were fore the end of the primary-task stimulus. For each subject, the MRT affected similarly (the interaction between age group and delay calculated in the firstbloc k was used to set the six stop-signal delays for for the probability of inhibition was nonsignificant, F(15, the second block equal to MRT - 500 ms, MRT - 400 ms, MRT - 300 ms, MRT - 200 ms, MRT -100 ms, and MRT - 0 ms. MRTs for trials 220) = 1.29. Moreover, the slope of inhibition functions, F(3, without stop signals in the second block were used to set the delays for 44) = 1.00, ns, did not vary among age groups (Table 1 and the third block, and MRTs for the third block were used to set the Figure 1). Although the adult and Grade 6 groups had shorter delays for the fourth, and so on. mean SSRTs than the younger age groups, the differences were Procedure. Adults were tested in the same way as children, except not significant, F(3, 44) = 2.4. that children were tested at school and adults at the hospital. Subjects Inhibition functions plotted against MRT minus stop-signal were tested individually. They were seated comfortably in front of the delay differ to some extent in height and slope (Figure 1, left computer screen in a quiet room and instructed to keep a fingero f the panel). The difference in height may be explained by differ- left hand on the left key and a fingero f the right hand on the right key ences in SSRT, whereas the difference in slope may be due to throughout the experiment. The experimenter remained in the room. differences in the variability of SSRT and of primary-task RT. Instructions for the choice response-time task were given first. Sub- The inhibition functions are even closer when plotted against jects were told to respond as quickly and accurately as possible. After the block of practice trials, they were told not to respond to the pri- ZRFT (Figure 1, right panel), which corrects for differences due mary-task stimulus whenever the beep occurred. They were also told to SSRT and due to primary-task variability. This finding sug- not to wait for the stop signal because it would not occur very often. It gests that some of the differences between age groups may have was explained that the stop signals would occur in such a way that been due to the slight differences in SSRT. sometimes subjects would be able to stop their response and some- Across all children, age correlated moderately with SSRT (r= times not. —.32, p < .05) but not with the probability of inhibition or with The following dependent variables were measured: reaction time to slope. IQ also correlated moderately with slope (r = .29, p < .05) the primary task for trials without stop signals, errors (pressing for X but not with the percentage of inhibition or with SSRT. when O was presented or vice versa), probability of response inhibition at each stop-signal delay, probability of nonresponse to primary-task trials without a stop signal, and SSRT. Discussion Statistical analysis. In order to examine the shape of each group's Developmental changes were evident in the performance in inhibition function, analysis of variance (ANOVA) with repeated mea- the stop-signal paradigm. Older children and adults responded sures across delay for each age group was performed on the probability more quickly to the primary task than did younger children, of inhibition. An interaction between group and delay would indicate differences in the shape of inhibition functions. These differences and they did so with less variability of RT and similar percent- were examined more precisely through a comparison of the slope of ages of error. the inhibition function for each group. Slopes were calculated by fit- Development does not significantly affect the ability to in- ting regression lines to the inhibition function of each subject. If group hibit a motor response, as shown by the effect of stop-signal differences were observed in the slope, regression lines were fitted to delay, the slope of the inhibition function, and the overall proba- 714 RUSSELL SCHACHAR AND GORDON D. LOGAN

Table 1 Performance on the Stop-Signal Paradigm in Experiment 1

Grade 2 Grade 4 Grade 6 Adults Newman-Keuls Measure M SD M SD M SD M SD F(3,44) P post hoc test MRT for correct responses 963 100 884 135 763 178 660 59 13.5 <.01 2 > 4 > 6 > adult Mean SD of MRT 240 40 201 40 154 49 109 31 23.6 <.01 2 > 4 > 6 > adult % of errors 8.4 7.2 7.8 4.4 7.0 7.8 3.3 2.1 1.9 ns %ofnonresponses 2.2 1.9 3.2 6.1 1.4 3.4 2.0 4.3 .4 ns Slope 9.9 3.4 9.2 4.1 10.3 3.7 12.0 5.1 1.0 ns Mean SSRT 326 93 276 81 253 49 264 54 2.4 ns Note. MRT = mean response time; SSRT = stop-signal response time. bility of inhibition. Younger children triggered the stopping sively and situationally hyperactive ADDH subjects were also process as frequently as older children and adults, and their distinguished because evidence suggests that their cognitive stopping processes did not differ significantly in speed. The function may be distinct (Chee, Logan, Schachar, Lindsay, & mechanism of inhibitory control thus seems well developed by Wachsmuth, 1989; Sandberg, Rutter, and Taylor, 1978). The Grade 2, although it may be less so in younger children. ADDH group was expected to show deficient inhibitory control The low rate of nonresponses to the primary task at all ages compared with the normal control (NC) group (i.e, to have a excludes the possibility that failure to inhibit simply reflected lower or flatter inhibition function). If an impulse control defi- distractibility or strategy (e.g, failing to attend to the target cit was specific to the ADDH group, this group's inhibition stimuli or trying to improve the chance of inhibiting by decid- function would also be lower or flatter than that of the patholog- ing before the trial not to respond regardless of the stimulus). ical control groups. These findings bring into question the apparent development in inhibitory control and the decrease in impulsivity with Method increasing age from the preschool years through childhood, changes detected with other tasks, for example, the MFFT (Ko- Subjects. The subjects in this experiment were 60 children referred gan, 1983). Such changes might result from improvement in for psychiatric or psychoeducational assessment of learning or behav- executive functions other than inhibitory control, increasing ioral disorders to the outpatient Departments of Psychiatry or Psychol- awareness of strategies, or improvement in attentional capacity ogy, The Hospital for Sick Children, Toronto, Canada, and 10 normal rather than from development in inhibitory control (see Case & volunteers recruited through an advertisement in a local home and school association newspaper. Subjects were compensated for their Globerson, 1974; Kahneman, 1973). travel expenses. Children were excluded from the study if they showed evidence of a neurological disorder such as epilepsy, had a history or Experiment 2: Childhood Psychopathology evidence of psychosis, or had an IQ of less than 80. All children had to be free of any medication for a minimum of 48 hours preceding test- The second experiment compared performance on the stop- ing. ping paradigm of groups of children with ADDH, conduct dis- Assessment of the child. The subjects were assigned a diagnosis on order (CD), learning disorder (LD), and emotional disorder the basis of a semistructured diagnostic interview with the parent(s), (ED) with that of normal children. It also included a group of behavior ratings completed by each child's classroom teacher, individ- children with both ADDH and CD because this presentation of ual assessment, and psychoeducational assessment. childhood hyperactivity is the most common (Stewart, De- Information was obtained by one of two child psychiatrists from Blois, & Cummings, 1980) and there is evidence that it might each child's parent(s) by following an interview protocol designed to have distinct correlates (Schachar & Wachsmuth, 1990). Perva- elicit symptoms relevant to diagnoses from the third edition of the DiagnosticandStatistical ManualofMental Disorders (DSM-III; Amer- ican Psychiatric Association, 1980). The interview covered prenatal, birth, postnatal, developmental, medical, academic, and family histo- Table 2 ries, as well as current child behavior and symptoms of psychopatho- Percentage Inhibition by Stop-Signal Delay in Experiment 1 logy. It also investigated the quality of the child's current interaction with peers, siblings, and adults. For each setting (e.g., at play out of Stop-signal delay Grade Grade Grade doors, in stores, with other children), parents were asked to describe a (ms) 2 4 6 Adults recent example of their child's behavior. Parents' subjective statements about behavior were not considered; instead, each symptom was rated MRT - 500 61 60 66 73 by the interviewer on the basis of its severity, age appropriateness, and MRT - 400 48 59 61 76 resultant degree of handicap. Only behaviors that were severe, handi- MRT - 300 47 46 48 64 capping, and age-inappropriate were considered diagnostic symp- MRT - 200 32 29 34 41 MRT - 100 18 28 26 30 toms. This interview was developed because we found that assessment MRT-0 12 18 18 22 based on a detailed description of child behavior rather than on questionnaires or structured interviews was necessary to distinguish symp- Note. MRT = mean response time. toms of hyperactivity from those of conduct disorder. Interrater reli- IMPULSIVITY AND INHIBITORY CONTROL 715 ability of this interview was assessed by having a second child psychia- of the following three criteria was considered evidence of clinically trist complete ratings of 18 audiotaped interviews. Raters agreed on the significant hyperactivity: (a) a rating of at least 5 out of 6 on the Rutter- presence or absence of 97% of symptoms (K = .92). In no case did B Hyperactivity factor, a score obtained by 3% of 10-year-old boys disagreement about individual symptoms result in disagreement (Schachar, Rutter, & Smith, 1981); (b) a rating of 4 Inattentive, 4 Impul- about the diagnosis. sive, and 3 Hyperactive items on the SNAP questionnaire, a score ob- The classroom teacher of each child completed the Rutter-B Rating tained by 5% of 10-year-old boys (Pelham et al, 1981); and (c) an abbre- Scale (Rutter, 1967), the ACTRS (Conners, 1973), and the SNAP ques- viated score of 15 or more on the ACTRS, a score predictive of a clinical diagnosis of hyperactivity (Goyette et al., 1978). Children who tionnaire (Pelham et al, 1981). Each child was interviewed by a psy- received both an interview- and a teacher-based diagnosis were consid- chologist following the format of Rutter and Graham (1968). In the ered pervasively hyperactive. However, from previous experience absence of accepted criteria for combining parent and child informa- (Chee et al, 1989), we expected to find that only 50% of the ADDH tion for research diagnoses, children were excluded from the study if group was pervasively hyperactive. The few cases of attention deficit the results of the child interview were markedly divergent from those disorder without hyperactivity were excluded from the study. of the parent interview. This occurred in the case of one child with pervasive developmental disorder. Children were assigned to the CD group if they met the DSM-HI Psychoeducational assessment. The psychoeducational assessment criteria for it, based on the parent-interview protocol. In addition, the diagnosis of CD was assigned to children diagnosed with oppositional consisted of four subtests from the WISC-R (Block Design, Object disorder (OD) whose symptoms were severe and pervasive, involving Assembly, Similarities, and Vocabulary; Wechsler, 1974) and of the relationships with both parents or a variety of adults. Of the 42 chil- WRAT-R (Jastak & Wilkinson, 1984). If these tests had been adminis- dren in the CD and ADDH + CD groups, 22 were diagnosed with CD tered to the child within the previous year, they were not given again. and 20 with OD. Of the 22 CD diagnoses, 14 were aggressive and 8 Child diagnosis. Diagnoses were assigned according to DSM-III nonaggressive. Typically, the aggressive children presented with long- criteria, and multiple diagnoses were permitted. Because a diagnosis standing symptoms of repetitive physical aggression toward peers and of ADDH does not require agreement between parents and teacher, of vandalism; the nonaggressive children showed a pattern of habitual ADDH could be diagnosed if either the parental interview or the lying and stealing. The majority of the OD group had more than the teacher report indicated its presence. An interview-based diagnosis of minimum number of symptoms (two) for the diagnosis and had opposi- ADDH was made if the parents reported three symptoms of inatten- tional symptoms that affected relationships with family and other tiveness, three of impulsivity, and two of hyperactivity in addition to a adults, as well as with peers and siblings. The practice of combining CD history of hyperactivity, impulsivity, and inattentiveness before the age and OD groups is consistent with the observation that the two diag- of 6 years. noses have not been differentiated and are qualitatively similar (An- A teacher-based diagnosis of ADDH was made if the teacher rated derson, Williams, McGee, & Silva, 1987; Reeves, Werry, Elkind, & the child as disturbed and significantly hyperactive. A rating of 9 or Zametkin, 1987; Werry, Methven, Fitzpatrick, & Dixon, 1983). more on the Rutter-B Rating Scale was taken as an indication of psychi- A diagnosis of ED was assigned to children meeting the DSM-III atric disturbance (Rutter et al, 1970), whereas the presence of any two criteria for any of the following: separation anxiety or overanxious,

0.8 0.8

0.0 co.e o £ £ 0.4 0.4 "5

0.2 10 0.2 O

-600 -400 -300 -200 -100 -1 -0.6 0 0.6 1 16 2 2.6 3 3.6 4 MRT - stop-signal daisy ZRFT Figure 1. Probability of inhibition of age groups by MRT minus stop-signal delay (left panel) and ZRFT (right panel). (MRT = mean response time; ZRFT = Z score, relative finishing time.) 716 RUSSELL SCHACHAR AND GORDON D. LOGAN

phobic, obsessive compulsive, or somatization disorders. LD was diag- ences in the probability of inhibition: The ADDH group inhib- nosed if the full-scale IQ was average or higher and the standard score ited fewer responses than the NC group, F(l, 64) = 6.97, p< .05, on the WRAT Reading subtest was below the 25th percentile in the but not fewer than the CD or ADDH + CD groups or all patho- absence of a physical or sensory disorder (Siegel & Heaven, 1986). logical groups taken together. On the basis of these criteria, the children were assigned to one of As predicted, the probability of inhibiting a response was five diagnostic groups: ADDH, CD, mixed ADDH + CD, ED, or LD. Multiple diagnoses were permitted. Children in the CD, ED, and LD strongly affected by the stop-signal delay, F(5,329) = 121.2, p < groups could not be hyperactive according to the criteria used, but .001. The probability fell from about 60% at the longest delays those in the CD and ED groups could have an additional diagnosis of to about 15% at the shortest (Table 4 and Figure 2, left panel). LD. Children with ADDH, ADDH + CD, or CD could have an addi- Variation in the stop-signal delay affected the probability of tional diagnosis of ED or LD. Volunteers qualified as controls (NC) inhibition of the diagnostic groups differently, as is evident in a only if they were free of any diagnosis. significant interaction between group and delay, F(25, 320) = The child diagnostic interview administered to parents did not in- 2.4, p < .01. Post hoc comparisons revealed that the effect of clude two of the new criteria for oppositional defiant disorder con- delay was significantly less in the ADDH group than in any tained in the revised edition of the DSM-III (DSM-IH-R; American other group: F(5, 320) = 9.7, p < .001, for the NC group; F(5, Psychiatric Association, 1987) but did include all items necessary for 320) = 5.3, p < .001, for the pathological groups considered revised diagnoses of ADDH, CD, and ED. All subjects in the latter together; F(5,320) = 5.0, p < .001, for the ADDH + CD group; three groups met criteria for the equivalent DSM-IH-R diagnosis. Al- and F(5, 320) = 2.9, p < .05, for the CD group. though not every item for oppositional defiant disorder was included, it was our clinical impression that all OD subjects would merit a DSM- The smaller effect of delay on the probability of inhibition in HI-R diagnosis of oppositional defiant disorder. the ADDH group was also evident in the analysis of the slopes Stopping task. The task used in Experiment 2 differed from that of the regression lines fitted to the probability of the inhibition used in Experiment 1 only by the addition of a block of practice trials. data (Table 3 and Figure 2), indicating that the ADDH group This additional practice block consisted of practice on the primary was affected by delay less than the other groups, F(5,64) = 3.0, task, as did the practice block in Experiment 1, but no stop signals were p < .05. Post hoc comparisons indicated that the slope of presented. In Experiment 2, the second practice block presented sub- the ADDH group's results was less than that of the NC group jects with both the primary task and stop signals to provide practice in (p<.05). stopping. These two practice blocks were followed by nine experimental blocks, as in Experiment 1. The MRT for each block was based on The probability of inhibition is plotted against ZRFT in the the MRT to the primary task of the preceding block. Stimuli were right panel of Figure 2, correcting for differences in primary- presented on the same computer and in the same way for both experi- task variability and SSRT. In this case, the ZRFT correction ments, and stop signals were randomized across conditions in the same appears to enhance the differences between the groups' inhibi- way. The experimental procedure was identical except that all testing tion functions. The ADDH group inhibited much less often was conducted at the hospital while the child's parents were being than did clinical controls at equivalent values of ZRFT. The interviewed. differences in slope between ADDH subjects and normal con- Statistical analysis. The analyses were conducted in the same way as trols seem less pronounced than when they are plotted against for Experiment 1. Because differences in inhibition were not expected MRT minus the stop-signal delay, but they remain apparent. between all of the groups, planned comparisons were conducted to These conclusions were confirmed by ANOVA. Group differ- compare the ADDH group with the NC group and each of the other groups (Kirk, 1982 pp. 832-839). ences in the slope of the inhibition functions plotted against ZRFT did not reach conventional levels of significance, F(5, 64) = 1.49, although the slope of the ADDH group remained Results much flatter than that of the other groups. This observation was confirmed by the planned comparisons indicating that the The diagnostic groups did not differ significantly in mean ADDH group had a flatter slope than either the NC, F(l, 64) = age (Table 3) or in mean IQ, although the NC group had some- 5.2, p < .05, or the ADDH + CD, F(l, 64) = 4.9, p < .05, groups. what higher mean IQs than the other groups. These slope differences indicate that greater variation in the Primary task processes. The diagnostic groups did not primary-task RT cannot completely explain the observed defi- differ in MRT to the primary task (see Table 3). However, cit in inhibitory control of the ADDH group. groups differed significantly in the variability of MRT. The Deficient inhibitory control was also evident in the differ- ADDH group showed the greatest amount of variability in ences in SSRT among the groups. The mean SSRT of the NC MRT and the ED group the least, but the intergroup differ- group was the shortest and that of the ADDH group the lon- ences tested post hoc did not reach conventional levels of signif- gest. The main effect of group did not reach conventional levels icance. The ADDH group made more errors on the primary of significance, F(5, 62) = 2.3, p = .06, but planned compari- task, but the difference was not significant. The rate of nonre- sons indicated that mean SSRT was longer for the ADDH sponse did not differ among diagnostic groups. group than for the NC group, F(l, 62) = 9.1, p < .01. However, Stopping processes. Although the mean probability of in- these differences in SSRT were insufficient to account for hibiting a response after a stop signal varied from 48% in the group differences in inhibition functions because these func- NC group to 28% in the ADDH group (Table 4), it did not tions were not aligned when plotted against ZRFT (i.e, there differ significantly among the groups, as is evident in the ab- were still large intercept differences). sence of a main effect for group in an ANOVA with factors for Across all subjects, age correlated weakly with the percentage diagnosis and stop-signal delay, F(5, 64) = 1.5, ns. However, of inhibition (r = .20) and the slope (r = .22) and moderately with planned comparison indicated the expected intergroup differ- the mean SSRT (r = -.29, p < .05). IQ was also weakly asso- IMPULSIVITY AND INHIBITORY CONTROL 717

Table 3 Performance on the Primary Task of Subjects in Each Diagnostic Group in Experiment 2

ADDH + NC LD ED CD CD ADDH

Measure M SD M SD M SD M SD M SD M SD F(5, 64) P Age (months) 120 14 120 14 119 16 117 23 112 17 111 21 .6 ns IQ 123 16 112 18 107 12 110 17 104 12 107 14 2.3 ns MRT for correct responses 901 131 889 117 845 146 920 146 952 163 901 213 .7 ns Mean SD of MRT 194 62 196 53 188 51 231 35 235 74 255 78 2.4 % of errors 8.9 9.4 7.1 3.6 7.2 4.9 9.2 5.9 6.7 6.2 15.3 11.8 2.3 ns % of nonresponse 1.3 1.3 1.2 1.5 1.8 2.0 2.1 1.3 2.4 1.9 3.1 3.3 1.4 ns Slope" 13.7 6.8 10.1 6.3 9.3 6.0 9.4 3.9 10.4 4.9 5.2 3.7 3.0 <.05 ZRFT slope 23.9 12.6 18.5 10.2 17.8 12.9 21.5 10.8 22.7 10.0 12.9 12.4 1.5 ns Mean SSRT 269 142 309 94 297 98 322 75 328 145 437 183 2.3 ns Note. NC = normal controls (n = 10); LD = learning disorder (n = 11); ED = emotional disorder (n = 13); CD = conduct disorder (n = 9); ADDH = attention deficit disorder with hy peractivity (n = 13; for ADDH + CD, n = 14); MRT = mean response time; ZRFT = Z score, relative finishing time; SSRT = stop-signal response time. * A Newman-Keuls post hoc test for slope showed ADDH < NC.

ciated with the percentage of inhibition (17) and the mean met the criteria both at home and at school. The performance SSRT (-.03), but it correlated moderately with the slope (r= .34, on the stopping task of these two subgroups was quite different. p<.01). Pervasively hyperactive children inhibited on only 13% of stop- IQ and inhibitory control. Even though the diagnostic signal trials averaged across all delays, in contrast to 38% for the groups did not differ in IQ, the possible confounding effect of situationally hyperactive group, F(l, 11) = 9.1, p < .05, and had IQ on group differences in inhibition was investigated because a significantly flatter inhibition function with a slope of 2.4, of the significant correlation between IQ and slope. The inter- compared to that of 6.9 for the situationally hyperactive sub- action between the covariate IQ and group was not significant, group, F(l, 11) = 6.0, p < .05. When the 5 pervasively hyperac- F(5,58) =1.7, indicating that the relationship of IQ and slope is tive subjects were excluded from group comparison, intergroup similar across groups (Kerlinger& Pedhazur, 1973). In an analy- differences in slope were no longer significant, .F(5,59) = 1.41. sis of covariance with IQ as covariate, the intergroup differences in slope were significant, F(5,63) = 75.6, p < .05, whereas the effect of IQ was not, F(l, 63) =2.17. Covariation for IQ Discussion reduced the difference between ADDH- and NC-adjusted Our findings indicate that these diagnostic groups do not group means, but the difference remained significant (Ker- differ in the executive processes that determine performance linger & Pedhazur, 1973). Adjustment for IQ did not greatly on the primary task. And, with the exception of the ADDH alter the slopes of the results of the pathological controls; these group, all subjects controlled their actions closely and inhibited groups had IQ-adjusted slopes that did not differ significantly them on command with great speed. Differences in inhibition from those of the NC or ADDH groups. functions remained after the ZRFT correction, indicating that Situational versus pervasive hyperactivity. As expected, only the inhibition mechanism of the ADDH group was triggered 5 of 13 children meeting the criteria for a diagnosis of ADDH less frequently or was substantially more variable, as well as slower. Moreover, intergroup variation in IQ did not account for differences in inhibitory control. These results support the validity of the narrowly defined Table 4 syndrome of pervasive hyperactivity rather than that of the Percentage Inhibition by Stop-Signal Delay in Experiment 2 more broadly defined ADDH diagnosis (see also Chee et al, Diagnostic group 1989; Sandberg et al, 1978; Schachar et al, 1981; Taylor et al, 1986). At the same time, deficient inhibitory control cannot Stop-signal delay (ms) NC LD ED CD ADDH + CD ADDH simply be attributed to the severity of disturbance: The mixed ADDH + CD group had the greatest number of symptoms of MRT - 500 76 58 59 60 61 43 psychopathology yet did not have the most severe deficit in MRT-400 70 53 53 58 61 36 MRT - 300 62 44 44 34 50 29 inhibitory control. MRT - 200 43 27 30 34 34 21 This is the second experiment in which we have observed that MRT - 100 22 20 23 22 22 25 the ADDH + CD group does not demonstrate the same cogni- MRT-0 13 11 16 16 14 16 tive deficit as the ADDH group (Chee et al, 1989). Differences Note. NC = normal controls; LD = learning disorder; ED = emotional between these two subgroups have also appeared in studies of disorder; CD = conduct disorder; ADDH = attention deficit disorder their clinical features. The mixed group is associated with a with hyperactivity; MRT = mean response time. greater incidence of psychosocial adversity and parental psy- 718 RUSSELL SCHACHAR AND GORDON D. LOGAN

0.8 0.8

0.6 0.6 O

— 0.4 0.4

0.2 0.2 2 0.

-600 -400 -300 -200 -100 -1 -0.6 0 0.6 1 16 2 2.6 3 3.6 MRT - 8top-8lgnal delay ZRFT Figure 2. Probability of inhibition of diagnostic groups by MRT minus stop-signal delay (left panel) and ZRFT (right panel). (NC = normal controls; LD = learning disability; ED = emotional disorder; CD = conduct disorder; ADDH = attention deficit disorder with hyperactivity; MRT = mean response time; ZRFT = Z score, relative finishing time.) chopathology (Biederman, Munir, & Knee, 1987; Schachar & MRT. MRT may reflect strategy or attentional capacity rather Wachsmuth, 1990; Szatmari, Offord, & Boyle, 1989) and a than inhibitory control. worse prognosis (Schachar et al, 1981). Taken together, these The analysis of the performance of normal and disturbed studies support the contention that the hyperactivity of chil- subjects in the stop-signal paradigm helps us to understand the dren with a diagnosis of CD may be qualitatively different from development of the executive function of inhibitory control in the hyperactivity of children without it. The former might repre- normal children and confirms the central role of deficient im- sent a correlate of psychosocial disturbance, whereas the latter pulse control in the pervasively hyperactive subgroup of ADDH might reflect a cognitive disturbance. subjects. Contrary to what previous research might have predicted, the inhibitory control of younger children did not differ significantly from that of older children. The observed improve- General Discussion ment with increasing age in the primary-task process in Experi- The results of these experiments confirm the predictions of ment 1 suggests that the executive functions involved in the the model of inhibitory control on which the stop-signal para- selection, execution, or maintenance of an optimal response digm is based (Logan & Cowan, 1984) and reflect the utility of strategy develop with age. Improvement in these functions the paradigm as a measure of inhibitory control. The paradigm rather than in inhibitory control may have led to the improve- explains impulsivity in terms of underlying processes rather ment in performance on tasks such as the MFFT. than describing only surface features. It allows researchers to The stopping task differs from tasks requiring impulse con- measure inhibitory control directly by analyzing inhibition trol that a child might confront in everyday life. The stop signal functions and estimating stop-signal reaction time. is salient and easily detected. Real-world stop signals may be As the model predicted, variation in the delay between the harder to discriminate. Consequently, inhibitory control could occurrence of the stop signal and the subject's MRT affected develop as a consequence of development in discrimination. the probability of inhibition in both studies. With longer de- Nevertheless, these results indicate that the ability to inhibit lays, the probability increased. The finding of Experiment 1 action after the signal is detected is well developed by Grade 2 that development in the stopping and primary-task processes and does not develop much beyond that. can occur independently also confirms the model. This model The stop-signal paradigm also shows that a specific deficit in assumes that the two processes (stopping and response) are in- inhibitory control might underlie the impulsivity of pervasively dependent and do not compete for resources. Moreover, the hyperactive children. This deficit may be evident in a range of dynamic tracking, which is part of the stopping paradigm, al- behaviors, such as failure to stop, look, and listen; failure to lows for the assessment of inhibitory control independently of inhibit attention to distracting stimuli; and proneness to acci- IMPULSIVITY AND INHIBITORY CONTROL 719 dents. The central role of deficient inhibitory control was also Conners, C. K. (1973). Rating scales in drug studies with children. evident in another study: We found that improvements in behav- Psychopharmacy Bulletin [Special issue: Pharmacotherapy with ior and academic performance resulting from treatment with children], 24-29. methylphenidate were strongly associated with improvement in Douglas, V I. (1983). Attentional and cognitive problems. In M. Rutter inhibitory control as measured by the stop-signal paradigm (Ed.), Developmental neuropsychiatry (pp. 280-329). New York: (Tannock, Schachar, Carr, Chajczyk, & Logan, 1989). Com- Guilford Press. pared with a placebo, treatment with methylphenidate was as- Douglas, V I, & Peters, K. G. (1979). Toward a clearer definition of the sociated with the increased probability of inhibiting a response, attentional deficit of hyperactive children. In G. A. Hale and M. Lewis (Eds.), Attention and cognitive development (pp. 173-248). increased slope of inhibition functions, and decreased SSRT. New York: Plenum Press. The stopping paradigm contrasts with the MFFT and other Goyette, C. H., Conners, C. K., & Ulrich, R. F. (1978). Normative data measures of impulsivity in which performance depends on fac- on revised Conners Parent and Teacher Rating Scales. Journal of tors such as MRT and strategy, as well as on inhibitory control. Abnormal Child Psychology, 6, 221-236. In addition, the probability of inhibition on the stopping task is Jastak, S., & Wilkinson, G. S. (1984). The Wide Range Achievement not as strongly correlated with IQ as is the performance on Test: Administration manual (rev). Wilmington, DE: Jastak Asso- other popular measures of impulsivity such as the MFFT, Por- ciates. teus mazes, or draw-a-line-slowly tasks (Milich & Kramer, Kagan, J, Rosman, B. L., Day, D, Albert, J, & Phillips, W (1964). 1984). Furthermore, in the study of Tannock et al. (1989), the Information processing in the child: Significance of analytic and effects of methylphenidate on MFFT performance were less reflective attitudes. Psychological Monographs, 78(1, Whole No. clear than its effects on stop-signal paradigm performance, sug- 578). gesting that MFFT latency and error measures are less sensitive Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall. to medication effects than are measures derived from the stop- Kerlinger, F. N., & Pedhazur, E. J. (1973). Multiple regression in behav- signal paradigm. ioral research. New York: Holt, Rinehart & Winston. Whereas these observations bear on the inhibition of discrete Kirk, R. (1982). Experimental design: Procedures for behavioural motor actions, research with adults indicates remarkable con- sciences. Pacific Grove, CA: Brooks/Cole. sistency in the degree of control across a range of motor, speech, Kogan, N. (1983). Stylistic variation in childhood and adolescence: and cognitive tasks (Logan & Cowan, 1984; Logan, 1985). Fur- Creativity, metaphor, and cognitive style. In P. Mussen (Ed.), Hand- ther research will be necessary to determine whether the prop- book of child psychology.Vol. 2. Cognitive development (pp. 630-706). erties of control are as consistent in children. However, it seems New York: Wiley. reasonable to assume that these results can be generalized to Logan, G. D. (1985). Executive control of thought and action. Ada other tasks and situations. Psychologia, 60,193-210. Logan, G. D, & Cowan, W B. (1984). On the ability to inhibit thought and action: A theory of an act of control. Psychological Review, 91, References 295-327. Logan, G. D, Cowan, W B., & Davis, K. A. (1984). On the ability to American Psychiatric Association. (1980). Diagnostic and statistical inhibit simple and choice reaction time responses: A model and a manual of mental disorders (3rd ed.). Washington, DC: Author. method.. Journal of Experimental Psychology: Human Perception and American Psychiatric Association. (1987). Diagnostic and statistical Performance, 10, 276-291. manual of mental disorders (3rd ed., rev). Washington, DC: Author. Luce, R. D. (1986). Response times. New York: Oxford University Press. Anderson, J. C, Williams, S., McGee, R., & Silva, P. A. (1987). DSM-III Milich, R., & Kramer, J. (1984). Reflections on impulsivity: An empiri- disorders in preadolescent children: Prevalence in a large sample cal investigation of impulsivity as a construct. Advances in Learning from the general population. Archives of General Psychiatry, 44, 69- and Behavioural Disabilities, 3, 57-94. 76. Pelham, W E, Atkins, M, & Murphy, H. A. (1981, September). ADD Ault, R. L., Crawford, D. E., & Jeffrey, W E. (1972). Visual scanning with and without hyperactivity: Parent, teacher, and peer rating cor- strategies of reflective, impulsive, fast-accurate, and slow-accurate relates. In W E. Pelham (Chair), DSM-III category of attention deficit children on the Matching Familiar Figures test. Child Development, disorder: Rationale, operation, and correlates. Symposium conducted 43,1412-1417. at the 89th Annual Convention of the American Psychological Asso- Biederman, J., Munir, K, & Knee, D. (1987). Conduct and oppositional ciation, Los Angeles. disorder in clinically referred children with attention deficit dis- Peloquin, L. J., & Klorman, R. (1986). Effects of methylphenidate on order: A controlled family study. Journal of the American Academy of normal children's mood, event-related potentials, and performance Child and Adolescent Psychiatry, 26, 724-727. in memory scanning and vigilance. Journal of Abnormal Psychology, Brown, A. L., Bransford, J. D, Ferrara, R. A., & Campione, J. C. (1983). 95, 88-98. Learning, remembering, and understanding. In P. H. Mussen (Ed.), Reeves, J. C, Werry, J. S, Elkind, G. S., & Zametkin, A. (1987). Atten- Handbook ofchild psychology (Vol. 3, pp. 77-166). New York: Wiley. tion deficit, conduct oppositional, and anxiety disorders in children: Campbell, S. B, Douglas, V I, & Morgenstern, G. (1971). Cognitive II. Clinical characteristics. Journal of the American Academy of styles in hyperactive children and the effect of methylphenidate. Child and Adolescent Psychiatry, 26,144-155. Journal of Child Psychology and Psychiatry, 12, 55-67. Rutter, M. (1967). A children's behaviour questionnaire for completion Case, R., & Globerson, T. (1974). Field independence and central com- by teachers: Preliminary findings. Journal of Child Psychology and puting space. Child Development, 45, 772-778. Psychiatry, 8,1-11. Chee, P, Logan, G, Schachar, R., Lindsay, P., & Wachsmuth, R. (1989). Rutter, M., & Graham, P. (1968). The reliability and validity of the Effects of event rate and display time on sustained attention in hy- psychiatric assessment of the child: I. Interview with the child. Brit- peractive, normal, and control children. Journal of Abnormal Child ish Journal of Psychiatry, 114, 563-579. Psychology, 17, 371-391. Rutter, M., Tizard, J., & Whitmore, K. (1970). Education, health and 720 RUSSELL SCHACHAR AND GORDON D. LOGAN

behaviour: Psychological and medical study of child development. Szatmari, P, Offord, D, & Boyle, M. (1989). Ontario Child Health New York: Wiley. Study: Prevalence of attention deficit disorder with hyperactivity. Salkind, N. J., & Wright, J. C. (1977). The development of reflection- Journal of Child Psychology and Psychiatry, 30, 219-230. impulsivity and cognitive efficiency. Human Development, 20, Yll- Tannock, R., Schachar, R. 1, Carr, R. P, Chajczyk, D, & Logan, G. D. 387. (1989). Effects of methylphenidate on inhibitory control in hyperac- Sandberg, S. T, Rutter, M., & Taylor, E. (1978). Hyperkinetic disorder tive children. Journal of Abnormal Child Psychology, 17, 473-491. in psychiatric clinic attenders. Developmental Medical Child Neurol- Taylor, E, Everitt, B., Thorley, G., Schachar, R, Rutter, M, & Wiesel- ogy, 20, 279-299. berg, M. (1986). Conduct disorder and hyperactivity: II. A cluster Schachar, R., Rutter, M., & Smith, A. (1981). The characteristics of analytic approach to the identification of a behavioural syndrome. situationally and pervasively hyperactive children: Implications for British Journal of Psychiatry, 149, 768-777. syndrome definition. Journal of Child Psychology and Psychiatry, 22, Wechsler, D. (1974). Wechsler Intelligence Scale for Children: Manual 375-392. (rev). New \brk: Psychological Corporation. Schachar, R., & Wachsmuth, R. (1990). Hyperactivity and parental Werry, J. S., Methven, R. J, Fitzpatrick, J, & Dixon, H. (1983). The psychopathology. Journal of Child Psychology and Psychiatry, 31, interrater reliability of DSM-III in children. Journal of Abnormal 381-392. Child Psychology, 11, 341-354. Siegel, L. S, & Heaven, R. K. (1986). Categorization of learning disabil- Winer, B. J. (1971). Statistical procedures in experimental design. New ities. In S. Ceci (Ed.), Handbook of cognitive, social, neuropsychologi- York: McGraw-Hill. cal aspects of learning disabilities (Vol. 1, pp. 95-121). Hillsdale, NJ: Erlbaum. Stewart, M. A., DeBlois, C. S., & Cummings, C. (1980). Psychiatric Received October 30,1989 disorder in the parents of hyperactive boys and those with conduct Revision received March 22,1990 disorder. Journal of Child Psychology and Psychiatry, 21, 283-292. Accepted April 2,1990 •

Butcher, Geen, Hulse, and Salthouse Appointed New Editors, 1992-1997

The Publications and Communications Board of the American Psychological Association announces the appointments of James N. Butcher, University of Minnesota; Russell G. Geen, University of Missouri; Stewart H. Hulse, Johns Hopkins University; and Timothy Salthouse, Georgia Institute of Technology as editors ofPsychological Assessment: A Journal ofConsulting and Clinical Psychology, the Personality Processes and Individual Differences section of the Journal of Personality and Social Psychology, the Journal of Experimental Psychology: Animal Behavior Processes, and Psychology and Aging, respectively. As of January 1,1991, manuscripts should be directed as follows:

• For Psychological Assessment send manuscripts to James N. Butcher, Department of Psychol- ogy, Elliott Hall, University of Minnesota, 75 East River Road, Minneapolis, Minnesota 55455. • For JPSP: Personality send manuscripts to Russell G. Geen, Department of Psychology, University of Missouri, Columbia, Missouri 65211. • For JEP: Animal send manuscripts to Stewart H. Hulse, Johns Hopkins University, Depart- ment of Psychology, Ames Hall, Baltimore, Maryland 21218. • For Psychology and Aging send manuscripts to Timothy Salthouse, Georgia Institute of Technology, School of Psychology, Atlanta, Georgia 30332. Manuscript submission patterns make the precise date of completion of 1991 volumes uncer- tain. Current editors will receive and consider manuscripts through December 1990. Should any 1991 volume be completed before that date, manuscripts will be redirected to the newly appointed editor-elect for consideration in the 1992 volume.